Newborn screening for cytomegalovirus

Transcription

1 Newborn screening for cytomegalovirus External review against programme appraisal criteria for the UK National Screening Committee (UK NSC) Version: Final Bazian Ltd. October 2017 This analysis has been produced by Bazian Ltd for the UK National Screening Committee. Bazian Ltd has taken care in the preparation of this report, but makes no warranty as to its accuracy and will not be liable to any person relying on or using it for any purpose. The UK NSC advises Ministers and the NHS in all four UK countries about all aspects of screening policy. Its policies are reviewed on a 3 yearly cycle. Current policies and the policy review process can be found at

2 Abbreviations List ABR AUC CI (C)CMV CNS CSF CT DBS FN FP IUGR LOD LR LSV MRI NICU NPV OR PCR PPV RCT Sn SNHL Sp SR TN TP Auditory brainstem response Area under the curve Confidence interval (Congenital) Cytomegalovirus Central nervous system Cerebrospinal fluid Computed tomography Dried blood spot False negative False positive Intrauterine growth restriction Limit of Detection Likelihood ratio Lenticulostriated vasculopathy Magnetic resonance imaging Neonatal intensive care unit Negative predictive value Odd ratio Polymerase chain reaction Positive predictive value Randomised controlled trial Sensitivity Sensorineural hearing loss Specificity Systematic review True negative True positive Page 2

3 Plain English Summary Cytomegalovirus is a common viral infection in children and adults. In the UK around half the population is infected by it at some point in their life. However, in a healthy person it causes no or few symptoms, and the majority people do not need treatment. Cytomegalovirus infection spreads by contact of body fluids such as saliva, blood and urine Cytomegalovirus infection can cause problems in some newborn babies if they catch it from their mother during pregnancy. This is called congenital cytomegalovirus infection. Research carried out in the UK in the 1970s, suggested that around three in every thousand babies are born with congenital cytomegalovirus infection. That is around 2,400 babies each year in the UK. But, most of these babies will develop normally. Of all newborns that are born with the infection 10 to 15% will show some moderate or severe symptoms in the first two weeks of life (240 to 360 per year in the UK). Around half of them will develop adverse long term outcomes, such as hearing loss and other neurological problems (120 to 180 babies per year in the UK However, of the 85 to 90% of the babies with no clear symptoms at birth around 10% will later develop problems such as hearing loss. This means that, each year in the UK between 2,040 to 2,160 babies will be born with the condition, but will develop normally and 202 to 216 will have some problem later on. In 2012 the UK National Screening Committee (UK NSC) recommended against screening for cytomegalovirus. The current review looks at research published since then to see if the recommendation is still valid. In pregnancy screening is not recommended because: 1. There is still no reliable screening test to detect cytomegalovirus infection during pregnancy. 2. No treatment was identified that could prevent the developing baby in the womb getting cytomegalovirus infection from their mother. Newborn screening is not recommended because: 1. An option for newborn screening would be to test a saliva sample but research is needed to understand more about it. 2. No reliable way of knowing which babies are going to develop long-term health problems from cytomegalovirus infection. The review found some research looking in to this. It included looking at newborn blood test results or scanning the baby s brain. But this would also need more research to ensure that these approaches were reliable and safe. 3. Screening is likely to identify a greater number of infants with ccmv than is currently the case. These are likely to have minimal symptoms or no symptoms. The management and treatment approach for these children is unclear, and it is unknown whether screening improves their outcomes. Page 3

4 Executive Summary Aim of the review This review assesses whether newborn screening for congenital cytomegalovirus (CCMV) should be offered. The review considers literature published between January 2011 and February 2016, the date of the last UK NSC review. Background Cytomegalovirus (CMV) is a common viral infection of the herpes genus with estimated seroprevalence rates of around 50% * in the adult UK population. Though it causes no, or minimal, symptoms in healthy adults it can have serious consequences for fetuses/newborns exposed to infection during pregnancy. Research carried out in the 1970s suggested that around three in every thousand babies is born with congenital CMV. That is around 2,400 babies each year in the UK.An estimated 10-15% of newborns with CCMV (240 to 360 per year) have symptoms at birth including neurological and focal organ involvement (in particular spleen, liver, lungs, lymph nodes and central nervous system). Another 10-15% of asymptomatic newborns will develop late sequelae, mostly sensorineural hearing loss (SNHL). CCMV has been estimated to account for around one in five cases of SNHL. To minimise the impact of CCMV several strategies have been proposed by experts in the field.[1-6]these strategies include: earlier identification through maternal screening, vaccination, preventative and behavioural interventions, treatment for infected pregnant women, and screening and treatment of affected newborns. Previous/ Current UK NSC Review The UK NSC currently recommend against antenatal or newborn screening for CMV. This recommendation dates from the previous external review in 2012, which highlighted several uncertainties. Antenatal screening The uncertainties relating to antenatal screening included: a lack of clarity on the risks to the fetus associated with primary and non-primary infection screening and diagnostic strategies, because there is not a test or combination of tests that can predict outcome in the infant no interventions to prevent mother-to-child transmission or minimise the severity of infection Antenatal screening is only discussed briefly as part of the introduction to this review. This is because very little information on these issues was found in the literature search, and the focus of attention relating to screening has shifted to the newborn period. * This prevalence data is based on data from 1980s and 1990s and therefore may not account for recent demographic changes that may affect the underlying infection rates Page 4

5 Newborn screening The 2012 UK NSC review concluded that both newborn dried blood spots (DBS) and saliva swabs had potential for CCMV screening. DBS would be the obvious choice as the sample is routinely collected to screen for other conditions. However, this approach had not been demonstrated to be sensitive enough for use in a large-scale newborn screening programme. Only one treatment was available, six weeks of intravenous ganciclovir, which should only be use in symptomatic newborns with neurological manifestations. A trial was underway investigating an oral formulation, valganciclovir, as an alternative. No treatment was available for asymptomatic infants or those with transient or non-specific symptoms. The previous review also highlighted a need to more accurately define risk in newborns and identify specific diagnostic signs or markers that could predict which newborns were likely to develop long-term sequelae. The current review considers whether the volume and direction of the evidence produced since the 2012 review addresses these questions and indicates that the previous recommendation not to screen newborns for CCMV should be reconsidered. Findings of current review a. The main candidate for a newborn screening test appears to be polymerase chain reaction (PCR) evaluation of saliva samples. This approach was assessed by two cohort studies: One large US cohort including 73,239 newborns (reported by two publications), and an Irish cohort of 1044 newborns. Both of these studies assessed PCR assay of a one-off liquid or dried saliva sample. However, there were concerns about verification bias in both of these studies, as they did not perform confirmatory testing (saliva and urine re-testing) of the full study sample. Therefore test performance cannot be known with accuracy. They also have the limitation that they do not consider the test in the context of a diagnostic pathway and its ability to change the management of newborns found to have CCMV. b. There is still a lack of clarity about how to identify newborns that will develop long-term sequelae, and therefore may benefit from medical intervention. One UK guideline recommends treating newborns with central nervous system (CNS) involvement or severe focal organ involvement. However, these recommendations are drawn from the single RCT of intravenous gancilovir combined with expert opinion. The guideline authors acknowledged the need for large studies of predictive markers. Three small cohort studies assessed the potential of specific CNS signs or viral load to predict the likelihood of long-term sequelae. However, symptomatic definitions varied widely across these studies and it is difficult to know how relevant or applicable these potential predictive markers may be to a population of newborns with CCMV identified through universal screening. c. The treatment to be offered to babies with screen detected CCMV remains unclear. Page 5

6 One RCT assessed valganciclovir as oral alternative to ganciclovir in symptomatic newborns, comparing six weeks with six months of treatment. The trial found that six months treatment with valganciclovir did not have a statistically significant effect on the primary outcome (best-ear hearing at six months) compared to six weeks treatment. There was some evidence that valganciclovir had a moderate, statistically significant positive effect on longer-term hearing (total-ear hearing) and neurodevelopmental outcomes at 12 to 24 months. The study was small, and its relevance to a screendetected population was uncertain. In addition no evidence was identified to establish the safety or effectiveness of oral valganciclovir compared to intravenous ganciclovir in terms of severity of hearing impairment or other complications at birth. No studies have assessed treatment for asymptomatic newborns. d. There remains a lack of clarity whether screening improves long-term outcomes. One cohort demonstrated that among a group of symptomatic newborns diagnosed with CCMV during the first weeks of life, those who were tested on clinical suspicion had poorer childhood outcomes, including hearing loss, than those who were tested as part of routine screening. However, no information was provided on the management strategy or its implementation in either group. Therefore, the study cannot provide evidence that the lower rate of adverse outcomes in the screening group is the direct effect of screening. Screening recommendations based on the current review The findings of the this review indicate that the current recommendation not to perform universal newborn screening for CCMV should be maintained. Page 6

7 Introduction Cytomegalovirus (CMV) is a common viral infection of the herpes genus (called also human herpes virus 5, HHV-5). Previous estimates suggest that seropositivity rates among the general adult population in the UK are around 50%, CMV infection usually causes no or minimal flu-like symptoms in healthy adults, but it can have serious consequences for immunocompromised individuals, and newborns exposed to infection during pregnancy or, in the case of preterm infants, during the postnatal period. The 2012 UK NSC review on screening for CMV[7] during the antenatal and/or postnatal period considered CCMV an important health issue. It was reported that the birth prevalence of CCMV (CCMV) in Europe was around three to five per 1000, and in the UK was estimated to be three per 1000 in the late 1970s. This may have changed as a result of shifts in population characteristics over the last few decades, but there were no more recent estimates. The review stated that 10-15% of neonates with CCMV present with symptoms, including petechiae, hepatomegaly, splenomegaly, hepatitis, and/or neurological signs such as microcephaly, chorioretinitis and intracranial calcification. About half of these children develop permanent sequelae with adverse outcomes higher in those with neurological presentations at birth. A further 10-15% of those who are initially asymptomatic develop neurological manifestations, mainly sensorineural hearing loss (SNHL), which may be bilateral, moderate or severe, or unilateral. Congenital CMV was estimated to account for 15-20% of moderate to profound, permanent bilateral hearing loss, but the true burden of the disease within the UK was unknown. To minimise the impact of CCMV several approaches have been proposed by experts in the field. [1] [2-6] These approaches include: earlier identification of infection through antenatal screening, vaccination of pregnant women, preventative and behavioural interventions in pregnancy, treatment for newly infected pregnant women, and screening and treatment of affected newborns. The 2012 UK NSC review[7] considered universal screening in the antenatal and/or newborn periods. The conclusion was that: a. In pregnancy, there were uncertainties regarding natural history. Primary or new infection in the mother was thought to carry highest risk of transmission to the fetus, but there were uncertainties over the contribution of non-primary CMV infection (i.e. reactivation or reinfection in seropositive women) to the disease burden. There was a need to better understand how to refine the risk of adverse outcomes in the fetus/newborn depending on the type of maternal infection. Antenatal screening tests lacked sufficient sensitivity and no interventions had been shown to be effective in preventing acquisition of maternal infection or reducing the risk of transmission to the This prevalence data is based on data from 1980s and 1990s and therefore may not account for recent demographic changes that may affect the underlying infection rates Page 7

8 fetus. Although progress has been reported towards the development of a vaccine to prevent maternal primary CMV infection, more research was needed to evaluate this. b. In the newborn period, the tests available had not been shown to be sufficiently reliable for screening, in particular newborn dried blood spot (DBS), which would be the obvious platform for CMV screening as this is routinely taken to screen for other conditions. There was some evidence of moderate benefit from the available intravenous or oral antiviral therapies. Due to these uncertainties the UK National Screening Committee concluded that screening for CMV in pregnancy or in the newborn period was not recommended. Antenatal screening for cytomegalovirus Cytomegalovirus infection during pregnancy Congenital infection in the newborn can be acquired through transplacental transmission from mothers with primary or non-primary CMV. Primary CMV occurs in women who are seronegative at the time of conception and then acquire CMV infection during pregnancy. Nonprimary infection occurs when a woman who is already seropositive either has reactivation of the dormant virus during pregnancy or is infected by a different viral strain. The risk of transmission to the fetus is thought to be greater with primary than with non-primary maternal infection.[8, 9] Placental transmission has been estimated to occur in around one third of cases of primary infection, with just over one in 10 cases resulting in symptomatic congenital infection.[8] However, transmission to the fetus and the effects that this might have on the baby vary with gestational age at infection.[10, 11] The 2012 UK NSC review[7] stated: although early studies suggested no differences in rates of intrauterine transmission by trimester of maternal infection, there is increasing evidence that seroconversion in late pregnancy is associated with a higher rate of congenital infection. However, transmission later in pregnancy appears to be associated with a lower risk of damage to the fetus. There is less clarity over the risk of congenital infection and fetal sequelae following non-primary infection. There are no currently UK-wide figures available, but a study in one city in England suggested that the prevalence was around 50% in white women and around 90% or higher in women of South Asian origin. This is similar to the West London findings based on data collected in the late 1970s/early 1980s (where prevalence ranged from under 50% in white women to almost 90% in Asian women).[12, 13] The 2012 UK NSC review[7] noted evidence that socioeconomic status and ethnicity may have an influence on CMV seroprevalence. One study from the 1980s reported higher seropositive rates in older, parous women, and in those who were unmarried. Seropositivity was also reported to range from 46% in White women to 88% in women of Asian origin. Therefore, changes in the demographic profile of the UK population in recent decades, and the consequential changes in some practices such as breastfeeding, may have had an influence on seropositivity rates. Though the risk from non-primary maternal infection is uncertain, the 2012 review[7] also highlighted how geographic areas with high seropositive rates (in general, communities of lower socioeconomic status and developing countries) tend to have high prevalence of CCMV (above Page 8

9 1%). As such, non-primary infection could contribute more to the prevalence of the disease than previously thought. As part of the current review, a literature search and appraisal was first conducted to review new evidence published since the 2012 UK NSC review[7] regarding antenatal CMV. The aim was to see whether there was new evidence that better described the natural history of CMV in pregnancy (in particular the risks from primary vs. non-primary infection), and in addition whether there was new evidence on screening tests and antenatal treatment. The evidence found is summarised in the tables in Appendix B. Natural history of maternal infection No large studies were identified that clarify whether risk of fetal transmission or severity of outcomes differs between primary and non-primary infection. One Japanese single-centre cohort (Ebina et al. 2014,[14] Appendix B.i) provided some evidence to further support the understanding that risk of fetal transmission is higher with primary infection (as indicated by low IgG avidity index) than with non-primary infection. However, recent evidence was not able to clarify whether the risk of fetal/newborn complications differs between primary and non-primary infection. Two small cohort studies of children with CCMV (Yamamoto 2011[15] and Townsend 2013[16], Appendices ii and iii) found that more children with SNHL were born to mothers with non-primary infection. This again seemed to confirm the pattern found in previous reviews that in populations with high seropositivity, a greater number of CCMV complications may be attributed to non-primary infection. However, the size of these studies and other applicability issues restricted meaningful analysis. Meanwhile other systematic reviews (de Vries 2013[17] and Goderis 2014,[18] Appendices iv and v) suggested that the risk of SNHL may be more or less equivalent following both primary and non-primary infections. However, again, there were various quality limitations and potential sources of bias in the studies informing these reviews. Similarly, a limited body of evidence was also found that looked at the timing of onset of maternal primary infection. Again these confirmed the pattern of results from earlier reviews that transmission risk may be higher for infections acquired during the third trimester, but early pregnancy infections may be more likely to result in infant with CCMV related-symptoms (Appendix B.vi). Screening for maternal infection Maternal infection is diagnosed usually through CMV IgM or IgG serology. IgG avidity is an indicator of the strength with which IgG antibodies bind to the antigen. Avidity is low in the weeks following acute infection and then progressively increases. Therefore a low IgG avidity index result in combination with a positive CMV IgM antibody is indicative of infection within the preceding three months. However, despite the availability of IgG avidity testing, the test cannot confirm exact timing of infection, because the cut-off for low avidity is not well established. Considering the antenatal period, several serology based strategies to reduce the burden of CCMV have been suggested: Page 9

10 1. Universal screening of all women in early pregnancy: this approach would identify the seronegative women who could then be offered serial serology tests during pregnancy to identify seroconversions. 2. Performing one off serology on all pregnant women at around 20 weeks (including avidity testing) to identify primary early pregnancy infections considered to be more likely to result in symptomatic CCMV. 3. Testing only women at increased risk of primary infection, such as those with frequent or prolonged contact with young children (under three years of age), for example, women living with young children or working in a day-care setting. 4. Performing targeted assessment at the second trimester ultrasound for features of CCMV (such as ventriculomegaly, intracerebral calcifications, microcephaly, echogenic bowel, intra-uterine growth restriction), and secondary maternal serology screening if positive features are identified. These strategies are limited as they mainly focus on primary CMV infections. Antenatal screening strategies aimed at identifying seroconversions would by definition exclude women whose babies may be at risk of CCMV due to non-primary maternal infection. The search identified only a single prospective cohort study (Yoshida et al. 2013,[19] Appendix B.vii) which shared similarities with strategies 1 and 2 above, but was not an exact match to either. This study screened all women for CMV IgM during the first trimester, and those who screened positive then received testing for CMV DNA in the amniotic fluid. The study demonstrated that serological CMV IgM screening has an extremely low positive predictive value to indicate fetal infection (4.8%). Otherwise the study could give no further screening test performance data due to the lack of follow-up of screen negatives. Antenatal interventions In addition, the evidence on options for intervention in the antenatal period to prevent acquisition of infection, mother-to-child transmission or attenuation of complications in the newborn remains limited. Primary prevention: although behavioural interventions (as suggested by the NHS[20]) may be effective in improving hygienic practices and preventing acquisition of primary infection in pregnant women, the 2012 UK NSC review[7] found no evidence to support this and highlighted that such interventions had not been tested under controlled conditions. An RCT published in 2015 looked at the effectiveness of providing pregnant women at risk for primary CMV with hygiene information. This study found that women who received such information were less likely to seroconvert. Only 4/331 (1.2%) of women who received the intervention seroconverted compared to 24/315 (7.6%) in the comparison group. It also found that the majority women felt hygiene recommendations would be beneficial for all pregnant women at risk for infection.[21]. Secondary prevention: two interventions have been proposed for the prevention of CMV transmission or complications in the newborn following primary maternal infection. These are antiviral therapy and CMV hyperimmune globulin treatment. The 2012 UK NSC review[7] noted some evidence from a small study suggesting that CMV hyperimmune globulin was associated Page 10

11 with a reduced risk of fetal transmission and that this was being investigated in randomised controlled trials. The updated search identified a single phase II RCT, the results of which contrasted with this early finding. Revello et al. (2014)[22] (Appendix B.viii) found that hyperimmune globulin was ineffective compared with placebo at preventing transmission of maternal primary infection acquired during the 1 st or 2 nd trimester. However, the sample size was relatively small and the overall transmission rate low. The authors calculated that triple the sample size would have been needed to have sufficient power to detect a treatment effect. The RCT also provided no information on whether treatment prevents complications in the newborn. No evidence was found to inform whether hyperimmune globulin may have a different effect following nonprimary infection. Information on use of antiviral drugs for CMV infection in pregnancy to prevent transmission remains very limited. Conclusion regarding antenatal screening The 2012 UK NSC review[7] noted that due to the complexities around diagnosis of maternal and fetal CMV infection and the lack of available interventions to prevent transmission or development of CCMV disease, the focus of attention had shifted in recent years towards neonatal screening instead of antenatal screening. The updated literature search found no significant new evidence to suggest that the UK NSC should reconsider screening in the antenatal period. Evidence on natural history continues to suggest that primary maternal infection carries higher risk of fetal transmission, but that the role of non-primary infection in the burden of CCMV morbidity may be considerable due to high population seroprevalence rates. The optimum antenatal screening strategy remains unclear. The review identified no primary studies exploring the effectiveness or practicality of the strategies previously proposed in discussion papers. Evidence from a single RCT did not confirm the effectiveness of antenatal hyperimmune globulin suggested in previous studies. No further evidence was identified on primary or secondary preventative strategies. Therefore, the decision was made to focus the 2016 review on the postnatal period, with the aim to establish if the evidence produced since the previous review, is sufficient to gauge whether the UK NSC should revisit the 2012 recommendation about a national neonatal screening programme for CCMV. Newborn screening for cytomegalovirus Congenital infection in the newborn can occur with different levels of severity from asymptomatic with no signs or symptoms of the disease (including normal hearing), to moderate and severe manifestations with multiple signs or symptoms and central nervous system (CNS) involvement. Of all newborn that have CCMV 10 to 15 % of will have moderate to severe symptoms in the first two weeks of life (240 to 360 per year). Thrombocytopenia, petechiae, hepatomegaly, splenomegaly, intrauterine growth restriction and hepatitis (raised transaminases or bilirubin) Page 11

12 Around 85 to 90% of babies that have CCMV will not have no apparent symptoms at birth and only around 10% will manifest problems such as hearing loss later on[18] This group of babies is most likely to be the potential target and beneficiary of newborn screening. Conclusion from the 2012 UK NSC review The UK NSC currently recommends against newborn screening for CMV. This followed the previous external review in 2012,[7] which highlighted several key uncertainties including the lack of a sensitive newborn screening test and no clear evidence that newborn treatments are effective. The current review considers whether the volume and direction of the evidence produced since the 2012 review indicates that the previous recommendation should be reconsidered. The 2012 review[7] conclusions in relation to newborn screening were: a. Both neonatal dried blood spots (DBS) and saliva swabs had the potential of becoming strategies for CCMV newborn screening in the future. DBS would be the obvious platform as it is routinely used to screen for other conditions. However, this approach had not been demonstrated to be sensitive enough for use in a large-scale newborn screening programme. b. Intravenous ganciclovir was the only recommended treatment for infants with CCMV. However, it had only been tested in infants with neurological manifestations who were at risk of developing adverse sequelae. For asymptomatic infants or those with transient or non-specific symptoms, it was not possible to predict whether adverse outcomes would develop to make treatment worth the risk. An oral formulation, valganciclovir, was being investigated as an alternative to ganciclovir in a clinical trial of short- versus long-term treatment in infants with any symptoms (not restricted to neurological). Page 12

13 Current update review The current review considers whether the volume and direction of the evidence produced since the 2012 external review indicates that the previous recommendation should be reconsidered. Five main criteria will be considered, with particular focus given to areas the 2012 review identified as uncertain, or supported by insufficient evidence. The main criteria and key questions reviewed are: Table 1. Key questions for current CMV update review Criterion Key Questions (KQ) # KQ Studies Included 4. There should be a simple, safe, precise and validated screening test. 1. What is the performance of screening strategies for detecting CCMV infection in newborns using tests based on dried blood spot, saliva or urine samples? 1 SR and 2 primary studies 9. There should be an effective intervention for patients identified through screening, with evidence that intervention at a presymptomatic phase leads to better outcomes for the screened individual compared with usual care. Evidence relating to wider benefits of screening, for example those relating to family members, should be taken into account where available. However, where there is no prospect of benefit for the individual screened then the screening programme should not be further considered. 10. There should be agreed evidence based policies covering which individuals should be offered treatment and the appropriate treatment to be offered. 2. What is the effectiveness and safety of treatments for CCMV? Treatments of interest are: a) Ganciclovir b) Valganciclovir c) Combination therapy of the above d) [Any licensed treatment that can be offered to newborns can include off-label use of licensed treatments] 3. Is there evidence that treatment is effective in newborns with different signs and symptoms of CCMV? For example, children with bilateral or unilateral hearing impairment? 4. Has an evidence based pathway been identified which can distinguish babies that are likely to be adversely affected by CMV and that may benefit from treatment? 1 primary study 0 studies 1 guideline, 9 primary studies Page 13

14 11. There should be evidence from high quality randomised controlled trials that the screening programme is effective in reducing mortality or morbidity. Where screening is aimed solely at providing information to allow the person being screened to make an informed choice (such as Down s syndrome or cystic fibrosis carrier screening), there must be evidence from high quality trials that the test accurately measures risk. The information that is provided about the test and its outcome must be of value and readily understood by the individual being screened. 5. Is there evidence that screening for CCMV impacts on morbidity (e.g. hearing) outcomes? 1 study Each criterion was summarised as met, not met or uncertain by considering the results of the included studies in light of the volume, quality and consistency of the body of evidence. Several factors were assessed to determine the quality of the identified evidence, including study design and methodology, risk of bias, directness and applicability of the evidence. Factors that were determined to be pertinent to the quality of the body of evidence identified for each criterion are outlined in the results section as well as the comment section of the Appendix tables. For Criterion 4, quality assessment focused on four main domains: patient selection, the index test, the reference standard, and flow and timing of index test and reference standard. Each domain was assessed for risk of bias, and the first three domains were assessed for applicability to a potential UK screening programme population. Details of these assessments can be found in the comment section of the Appendix tables. A systematic literature search of three databases was carried out looking for studies published between January 2011 and 19 February The search strategy is detailed in the appendix. Overall, the search yielded 2239 references addressing CMV. Of these, 289 were assessed as being potentially relevant to antenatal or newborn CCMV and were further filtered at title and abstract level. A total of 68 were selected for full text appraisal. Each section below provides additional information on the results of the evidence selection process for the given criterion. Across all questions, we excluded studies not available in English language, conference abstracts, letters, editorials and other communications, grey literature, Page 14

15 and studies of any design with sample size of less than 20 people. There were also six publications where the full text could not be identified. Selection and appraisal of studies was predominantly undertaken by one reviewer, any queries were resolved through discussion with a second reviewer, or with the UK NSC evidence team. The review was checked using Bazian Ltd s quality assurance process. Appraisal against UK NSC Criteria These criteria are available online at 4. There should be a simple, safe, precise and validated screening test. Description of the previous UK NSC evidence review conclusion The 2012 UK NSC review[7] described how diagnosis of CCMV requires a sample to be collected within the first two weeks of life, as testing after this time would not be able to distinguish congenital from postnatally acquired infection. However, few cases of CCMV are currently diagnosed clinically because most infected newborns are either asymptomatic or have nonspecific symptoms. Newborns with CCMV shed large amounts of the virus in saliva and urine. Diagnosis can be confirmed by viral culture of these specimens, or by polymerase chain reaction (PCR) analysis, which is rapidly replacing culture as the diagnostic method of choice due to greater efficiency and sensitivity.[23] Isolation of the virus in a urine sample has often been regarded as the gold standard method,[23] though urine collection from newborns has practical difficulties, which may preclude this from being considered for universal screening purposes. Collection of saliva, meanwhile, is relatively easy and non-invasive, though it would still require an additional sample to be taken from the newborn. Dried blood spot (DBS) screening would be the most obvious platform for CCMV screening as this sample is already routinely collected as part of the UK Newborn Screening Programme. However, previous studies had demonstrated sensitivity of the DBS test to be low as the viral load in blood is much lower than in saliva and urine. Therefore the 2012 review considered that either DBS or saliva swab could have potential roles in universal newborn CCMV screening. Current UK NSC key question The current review aimed to assess whether new evidence has been published since the last review that has assessed the performance of universal screening strategies for CCMV (involving either DBS, saliva or urine sample) conducted in all newborns. Description of the evidence Overall, 118 studies were identified as potentially relevant to CMV screening during title and abstract sifting. These studies were reviewed in more depth and 22 were accessed at full text. Priority was given to prospective studies including large, unselected samples of newborns (e.g. consecutively enrolled newborns) who would be representative of universal screening practice, or to any systematic reviews of these cohorts. Page 15

16 Studies were excluded that performed screening or diagnostic testing of selective populations of newborns thought to be at high risk of CMV. This included studies testing only symptomatic newborns (including those with SNHL), those born to mothers with active CMV infection or diagnosed with HIV, or fetuses/infants with other indications such as neurological-imaging findings, restricted growth, low birthweight or prematurity. As the question concerned screening for CCMV, rather than CMV that may be acquired during the later postnatal period, only studies where testing was performed within the first 2-3 weeks of life were included. Of the reviewed studies, one systematic review and two prospective cohort studies (one covered by two publications) met eligibility criteria for universal newborn CCMV screening. Wang et al. (2015)[24] conducted a systematic review and meta-analysis of newborn CMV screening by PCR assay of dried blood spot (Table 2, Appendix 1). This review included the results of two prospective cohort studies of DBS screening that have been published since the 2012 UK NSC review. As the cohorts were pooled in this meta-analysis, the primary publications were not reviewed separately. Boppana et al. (2011)[25] was a prospective cohort study of newborn evaluating the usefulness of real-time PCR assay of dried or liquid saliva samples as a screening test (Table 3, Appendix 2). Further results of the continuation of this screening programme were subsequently published in a brief report by Pinninti[26] (Appendix 3). The second prospective cohort, by Waters et al. (2014)[27] (Table 4, Appendix 4), examined newborn screening by PCR assay of liquid saliva sample. Page 16

17 Results Table 2: Newborn CMV screening by PCR assay of DBS vs. PCR or culture of urine or saliva sample Study Population characteristics Screen test/ Reference test Meta-analysis Sensitivity (95% CI) Specificity (95% CI) NPV (95% CI) PPV (95% CI) Wang et al. (2015)[24] (Appendix 1) Systematic review with meta-analysis (Search date: 1990 to Jan 2014) 15 cohorts n=26,007; 583 CMV positive 7 prospective, 8 retrospective 6 studies of universal screening; 9 selection based on suspected or confirmed infection. Individual study sample size range 19 to 11,407 DBS PCR assay vs. viral isolation or PCR DNA detection in urine and/or saliva (Method of DNA extraction/pcr amplification and reference standard used variable across studies) Collection age: DBS within the first week of life; urine/saliva within the first three weeks of life. All studies (n=15) Prospective studies (n=7) Retrospective studies (n=8) High assay LOD 1500 copies/ml (n=4) Low assay LOD <1500 copies/ml (n=4) (0.812 to 0.872) (0.548 to 0.693) (0.918 to 0.965) (0.773 to 0.914) (0.534 to 0.658) (0.998 to 0.999) (0.999 to 1.000) (0.974 to 0.989) (0.960 to 0.994) (0.999 to 1.000) (0.972 to 0.997) NR NR NR NR (0.835 to 0.948) NR NR NR NR Large DBS area >25mm 2 (n=5) (0.792 to 0.914) (0.997 to 1.000) NR NR Small DBS area 25mm 2 (n=5) (0.557 to 0.702) (0.999 to 1.000) NR NR Abbreviations: DBS, dried blood spot; LOD, limit of detection; NPV, negative predictive value; NR, not reported; PCR, polymerase chain reaction; PPV positive predictive value

18 Table 3: Newborn CMV screening by PCR assay vs. culture of liquid/dried saliva sample Study Population characteristics Screen test/ Reference test Specimen Sensitivity (95% CI) Specificity (95% CI) NPV (95% CI) PPV (95% CI) Boppana et al. (2011)[25] (Appendix 2) Prospective cohort 7 US centres June 08 to Nov 09 n=34,989 n=177 CMV positive by index and/or reference standard* Unselected sample: 98% from well-baby nurseries, 2% NICU Real-time PCR assay vs. rapid viral culture of liquid/dry saliva* Collection age: 1.0+/-1.2 days Liquid saliva (n=17,662) Dried saliva (n=17,327) 100 (95.8 to 100) 97.4 (90.8 to 99.7) 99.9 (99.9 to 100) 99.9 (99.9 to 100) 100 (99.9 to 100) 99.9 (99.9 to 100) 91.4 (83.8 to 96.2) 90.2 (81.7 to 95.7) * 79/93 screen-positives by PCR and/or culture liquid saliva received confirmatory testing (PCR and culture of both saliva and urine): calculated PPV for PCR assay 91.1% and PPV for rapid culture 98.6%. 74/88 screen-positives by PCR and/or culture dry saliva received confirmatory testing (PCR and culture of both saliva and urine): calculated PPV for PCR assay 94.4% and PPV for rapid culture 100%. Abbreviations: Cl, confidence interval; NICU, neonatal intensive care unit; NPV, negative predictive value; PCR, polymerase chain reaction; PPV positive predictive value Table 4: Newborn CMV screening by PCR assay of saliva sample vs. confirmatory re-testing with PCR and culture of urine and saliva Study Population characteristics Screen / Reference test Sensitivity (95% CI) Specificity (95% CI) PPV NPV Waters et al. (2014)[27] (Appendix 4) Prospective cohort Single Centre, Ireland June 2011 to May 2012 n=1044 asymptomatic infants n=4 screen positives; n=2 confirmed positive by reference standard Excluded: symptomatic infants, gestation <35 weeks Real-time PCR assay of saliva sample vs. confirmatory retesting with PCR assay of both urine and saliva and serology Collection age: first week of life 100% (95% CI to ) 99.74% (95% CI to 99.93) Reviewer calculated: 2 TPs/(2TPs + 2FPs) = 50% Cannot be calculated: Only screenpositives received confirmatory testing. FNs unknown Abbreviations: Cl, confidence interval; NICU, NPV, Negative predictive value; PPV positive predictive value; PCR, polymerase chain reaction Page 18

19 Dried blood spot testing The systematic review by Wang et al.[24] (Table 2, Appendix 1) examined the performance of PCR assay of DBS. It confirmed the conclusions of the 2012 UK NSC review that the sensitivity of screening DBS specimens is limited. Compared with the standard diagnostic methods of viral culture or PCR assay of newborn saliva or urine, the specificity of PCR assay of DBS is very high, suggesting there would be few false positives. However, the sensitivity is low, particularly when considering the pooled results of the prospective studies only. The retrospective studies included in the review gave better performance results, but these may have the potential for selection bias, including a higher proportion of suspected or confirmed cases than may occur in a general population sample. There were other limitations to quality and applicability of the evidence. The review specified that all index and reference test samples had to be collected within the first three weeks of life. Such practice ensures that the infection detected is congenital. However, the individual cohorts varied in the length of DBS storage (from days to years in some retrospective studies) and method of DNA extraction and PCR amplification. These variations may affect screening test performance. For example, subgroup analysis revealed that sensitivity was significantly affected by the limit of DNA detection on PCR assay, and DBS surface area. The individual cohort sample sizes also varied widely from 19 to 11,407, which could considerably affect the reliability of test performance results as CCMV would be relatively rare in a general newborn sample. There was reported to be low risk of bias around the reference standard used across studies. However, blinding to the result of the corresponding index or reference standard when interpreting the alternative test was not reported. In fact for 2/15 studies it was reported that the reference standard was known when interpreting the DBS result. Therefore there was the potential for reviewer bias when interpreting the index test. None of the 15 included studies came from the UK. An early UK study has tested the diagnostic validity of a rapid, single tube nested PCR method for enhanced detection of CMV in DBS compared with standard real time PCR assay. However, this has not yet been tested in a nonselected screening population. It is not clear which method of PCR could be an option for population screening or whether the timing of the test at 5 to 8 days would be appropriate for CCMV screening and further study would be needed. [28] The systematic review reports that five of the cohorts found no significant difference in test performance between samples taken from asymptomatic and symptomatic newborns. The review does not assess the value of DBS testing in identifying newborns at different risk of adverse outcomes from CCMV. Otherwise there was no indication from this review of whether DBS screen test results correlate with the risk of complications from CCMV. That is, whether the test is more likely to detect newborns at risk of adverse outcomes.

20 Saliva testing Two prospective cohort studies assessed the diagnostic performance of PCR assay of saliva samples. The first was a US cohort by Boppana et al.[25] (Table 3, Appendix 2). It reported on the performance of real-time PCR assay of dried or liquid saliva specimens in the detection of CCMV compared with the standard clinical test, rapid culture of saliva specimens. The study included 34,989 newborns (less than two days old) tested between June 2008 to December A second briefer report (Pinninti et al.[26], Appendix 3) included data on the longer term continuation of this study through to March 2012, including a final sample of 73,239 newborns. Boppana et al.[25] aimed to establish if using real-time PCR assay on saliva samples could be an effective, high-throughput and convenient method suitable for population screening. The study reported that dried saliva samples had slightly a lower sensitivity (>97%) than liquid samples (100%) when both were compared with the standard clinical test, saliva rapid culture. However, there are limitations to this study: a. The potential for over-detection from the test, or its consequences, was not explored. As a proof-of-concept study, the evidence shows that the test is clinically valid in accurately identifying the presence of CCMV infection. However, the test cannot identify those babies that will suffer from adverse long term outcomes. b. The clinical utility of the test was not fully explored in the study. The way in which the test affects management of the identified babies is a key evidence requirement for an evaluation of its usefulness as a screening tool, but was beyond the objectives of the study. The authors state that follow-up of the positive cases is ongoing but do not provide details. c. There is some concern about verification bias in the study. The study was primarily assessing the value of PCR assay of saliva compared with saliva viral culture which is used as standard clinical test. However gold standard confirmatory testing with viral culture of both saliva and urine samples, was only performed for babies who tested positive by either saliva PCR or culture. This gold standard testing demonstrated some misclassification from both PCR and viral culture of saliva. The possibility that there may have also been misclassification among those who tested negative by both saliva PCR and culture was not explored. d. Generalisability of the test and applicability to the UK population is unclear. The study was conducted in the US and the sample was taken at two days of life. If implemented in the UK the screening would probably be performed alongside the bloodspot screening visit at five days of life. The accuracy of the test on samples taken at different time was not explored. The majority of the study population (63.4%) was from ethnic minority groups (Asian 3.9%, Black 23.7%, White Hispanic 36.7% and other 3.3%) and so not representative of the UK sociodemographic. The continuation of this study (reported by Pinninti et al.[26], Appendix 3) contained evidence suggesting that PCR assay of saliva was superior to viral culture. The limited quality of the reporting in this brief publication prevents firm conclusions being drawn. However, due to the apparent superiority of PCR, the researchers stopped performing viral culture for those who screened negative by PCR assay of saliva. This publication therefore adds to the concerns of verification bias raised in point c above. Page 20

21 While this US cohort does not definitively establish PCR assay of saliva samples as a valid screening test, they do suggest PCR assay of saliva, particularly dried sample, as a practical candidate worthy of evaluation in future studies as a potential screening test. The second cohort by Waters et al.[27] (Table 4, Appendix 4) provides little additional information in this regard. It initially aimed to assess the performance of one-off PCR assay of saliva or urine sample; however, due to feasibility issues around collecting urine, the study reverted to collecting saliva samples only. PCR assay of saliva was compared against the reference standard of confirmatory re-testing of both saliva and urine, along with serology for CMV viral load and CMV IgM. The study has several applicability and quality issues. It was a relatively small sample from a single Irish centre. The samples were all collected within the first week of life so are relevant to CCMV screening, but only asymptomatic and full-term infants were sampled. Therefore it may have excluded newborns at highest risk from complications of CCMV. Sensitivity and specificity of one-off PCR assay of saliva were reported at near 100%, but there were only four screen positives. Two of these were negative on confirmatory re-testing, which gives a poor PPV of one-off PCR assay of only 50%. However, the very low prevalence rate in this asymptomatic sample, which may also differ from the rest of Ireland or the UK, limits the value of the PPV. Similar to the Boppana et al.[25] study there was potential for verification bias. Confirmatory retesting of saliva and urine was only performed for screen-positives. The number of false negatives is unknown and so NPV could not be calculated. It is unclear how the researchers calculated sensitivity and specificity. If the full cohort had received confirmatory re-testing this would have given a better indication of the reliability of one-off PCR assay of saliva as a potential screen test. Similar to the Wang et al.[24] review of DBS screening, neither the Boppana or Waters cohorts assessing PCR assay of saliva samples measured longer term outcomes in screen positives. Therefore it is not possible to know how well positive screen results by saliva sample correlate with health outcomes in the infant. Summary: Criterion 4 not met. One systematic review provides further evidence that newborn dried blood spot sampling is not sufficiently sensitive to be a reliable screening test for CCMV when compared with diagnostic methods of PCR assay or viral culture of saliva or urine. PCR assay of saliva samples has been identified as a candidate for a newborn screening test. This was based on the evidence from two cohorts that addressed screening by PCR assay on a oneoff dried or liquid saliva sample taken from newborns. However, these studies have potential for verification bias as they do not perform confirmatory testing of the full cohort. Its value as a screening tool would needs to be evaluated further An overriding issue concerning both DBS and saliva testing is that even if these approaches are able to detect CCMV with sufficient reliability, none of the studies reviewed here have yet reported longer-term disease outcomes. It is therefore not known how screening test results correlate with the likelihood of adverse outcomes. Page 21